Stochastic star formation in early galaxies: JWST implications

Abstract

The star formation rate (SFR) in high redshift galaxies is expected to be time-variable due to competing physical processes. Such stochastic variability might boost the luminosity of galaxies, possibly explaining the over-abundance seen at $z\gtrsim 10$ by JWST. We aim at quantifying the amplitude and timescales of such variability, and identifying the key driving physical processes. We select 245 $z=7.7$ galaxies with stellar mass $5\times 10^{6}\lesssim M_\star/{\rm M}_\odot\lesssim 5\times 10^{10}$ from SERRA, a suite of high-resolution, radiation-hydrodynamic cosmological simulations. After fitting the average SFR trend, $\langle {\rm SFR} \rangle$, we quantify the time-dependent variation, $\delta(t) \equiv \log [\rm SFR/\langle {\rm SFR} \rangle]$ for each system, and perform a periodogram analysis to search for periodicity modulations. We find that $\delta(t)$ is distributed as a zero-mean Gaussian, with standard deviation $\sigma_\delta \simeq 0.24$ (corresponding to a UV magnitude s.d. $\sigma_{\rm UV} \simeq 0.61$) that is independent of $M_\star$. However, the modulation timescale increases with stellar mass: $t_\delta \sim (9, 50, 100)\, \rm Myr$ for $M_\star \sim (0.1, 1, 5)\times 10^9\, {\rm M}_\odot$, respectively. These timescales are imprinted on the SFR by different processes: (i) photoevaporation, (ii) supernova explosions, and (iii) cosmological accretion/merging dominating in low, intermediate, and high mass systems, respectively. The predicted SFR variations cannot account for the required $z\gtrsim 10$ UV luminosity function boost. Other processes, such as radiation-driven outflows clearing the dust, must then be invoked to explain the enhanced luminosity of super-early systems.